Total Incidents
3,949
Total Deaths
2,598
Total Kidnappings
1,976
Total Wounded
2,730
Total Affected
7,304
Although both national and international aid workers face serious threats, national staff bear a disproportionate burden of harm. The distribution of outcomes reveals that nationals are more likely to be killed in incidents compared to their international counterparts. This pattern was confirmed through statistical testing, reinforcing that nationals are not only more frequently targeted but also experience more lethal consequences. These findings highlight the urgent need to strengthen protections for national personnel, who often operate in higher-risk, front-line settings with less institutional support.
Security incidents are not evenly distributed across humanitarian organizations. International and national NGOs experience the highest number of incidents—reflecting their broad operational presence in high-risk environments—followed by UN agencies and Red Cross affiliates. A Kruskal-Wallis H test confirmed that these differences are statistically significant, suggesting that some organization types face consistently higher levels of threat exposure. These findings underscore the importance of tailoring security protocols and duty-of-care standards to organizational context.
This chart displays the top predictors of fatal incidents among aid workers, based on logistic regression coefficients.
Risk profiles are data-driven categories that group countries based on patterns in aid worker security incidents—including frequency, fatality rate, and kidnapping rate. Each profile reflects a different level and type of operational risk:
These profiles help humanitarian actors allocate resources, assess threats, and tailor safety strategies to different contexts.
This PCA projection reduces the country-level incident data into two principal components, revealing clusters based on total incidents, fatality rate, and kidnapping rate. The axes represent linear combinations of those features and preserve the most variance in the dataset, making this view interpretable in terms of overall country risk.